I'm sure people have noticed the large number of questions on this board asking how to linearize things, handle conditional constraints, how to do Big M, etc.

Hello folks, this is the 2nd decade of the 21st century. There is plenty of software to do Big M modeling to linearize things, handle conditional constraints, handle non-convex occurrence of certain operators, etc. So long as appropriate bounds are provided for the relevant variables, automated software should be able to automatically do most of these things, without human error, and with as good or better formulations than the typical non-expert could generate.

For example, indicator constraints in CPLEX and many optimization modeling systems. Implies command in YALMIP, as well as under the hood automatic introduction of binary variables to handle non-convex occurrences of max or min, among many other things.

Hopefully, the posters who jump on these questions, with the same answers, more or less, over and over again, won't mind being put out of business, or at least having it reduced.

asked 10 Oct '16, 12:53

Mark%20L%20Stone's gravatar image

Mark L Stone
447310
accept rate: 15%

edited 10 Oct '16, 19:57

3

Well, you're probably right here... but, then again, abt. 50-75% of my karma points are due to answering basically the same question over and over again. Keeping this in mind, clearly not a surprise: *thumbs down*

(10 Oct '16, 14:47) fbahr ♦
2

@Mark: I take it you are of the opinion that people would actually check the FAQ? :-)

(10 Oct '16, 17:10) Paul Rubin ♦♦
1

@Paul Rubin Of course not. But their question could be answered with a comment to read the FAQ. Actually, I suggest making a sticky pinned to the top of the board. Maybe you and @fbahr can jointly co-author it. If you do a good job, I'll upvote you, which will be a kind of a karma severance pay. One of you can post it, and the other answer it, that way you can both get karma severance pay. So you could first of all reference automated software/modeling systems, etc. Then you could give examples of Big M modeling - just draw one example from each type of question you have answered many times.

(10 Oct '16, 19:46) Mark L Stone

If there's a way to put a "sticky note" at the top, I'd vote for one saying "please try the search engine before asking your question". As for automated software for big M modeling, I'm not familiar with any myself, but if you'd care to start a page on it and label it "community wiki", I'd support that. (I wonder if we should have a list of "wiki" pages somewhere?)

(11 Oct '16, 11:11) Paul Rubin ♦♦

A class of Big M constraints can be modelled using CPLEX indicator variables.

(11 Oct '16, 12:56) crypto

There has only been one new question subsequent to this one. Here it is :) https://www.or-exchange.org/questions/14277/either-or-constraint Can we award a special 10K karma bonus to @Paul Rubin for his very prescient comment above?

(12 Oct '16, 10:24) Mark L Stone
showing 5 of 6 show 1 more comments

A halfway comprehensive "Can I Linearize It" survey would go on for 10 pages at least. I just gave a whole talk on the complexities of linearizing (or getting software to linearize) quadratic optimization problems with integer variables! So I think it would be great if someone would write such a survey but I am not surprised that so far no one has found the time to do it.

On the topic of software capabilities . . . Perhaps the OR Software Forums page could also be pinned to the top of the board to cut down on the number of product-specific help questions.

link

answered 18 Nov '16, 11:12

4er's gravatar image

4er
64127
accept rate: 0%

Here is a nice, but not exhaustive, write up of a lot of linearizations, indicator constraints, big M (not called as such), etc. "MIP formulations and linearizations Quick reference" by FICO, 20 June 2009 http://www.fico.com/en/node/8140?file=5125

(26 Nov '16, 09:51) Mark L Stone

I suggest making a sticky pinned to the top of the board. Maybe @Paul Rubin and @fbahr can jointly co-author it. If you do a good job, I'll upvote you, which will be a kind of a karma severance pay. One of you can post it, and the other answer it, that way you can both get karma severance pay.

You could first of all reference automated software/modeling systems, and Constraint Programming, etc. Then you could give examples of Big M modeling - just draw one example from each type of question you have answered many times.

I'm not the best person to write such an "article". I suggest inviting representatives of modeling systems/automated software to include a description of their relevant capabilities with links to further info. For instance, invite Johan Löfberg https://github.com/johanlofberg to submit a description of relevant YALMIP capabilities. Similarly for CPLEX (including CP Optimizer for Constraint Programming), AMPL, GAMS, and AIMMS, which all have relevant capabilities, and other modeling and software tools. Invite anyone to add a description of whatever other tool or software they provide or know about.

link

answered 11 Oct '16, 13:23

Mark%20L%20Stone's gravatar image

Mark L Stone
447310
accept rate: 15%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "Title")
  • image?![alt text](/path/img.jpg "Title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Tags:

×65
×21
×7

Asked: 10 Oct '16, 12:53

Seen: 2,735 times

Last updated: 26 Nov '16, 09:51

OR-Exchange! Your site for questions, answers, and announcements about operations research.